Market event identification based on latent response to market events

Abstract

A market event identification method, system, and computer program product, include defining a likelihood function that maps a likelihood of a purchase of a product during an event by customers having similar customer features and estimating a likelihood of an event-driven purchase by the customers for each of a plurality of product-event combinations using the defined likelihood function.

What is claimed is: 
     
       1. A computer-implemented market event identification method, the method comprising:
 defining a likelihood function that maps a likelihood of a purchase of a product during an event by customers having similar customer features, the function including a variable that factors event-product relationships; 
 estimating a likelihood of an event-driven purchase by the customers for each of a plurality of product-event combinations using the defined likelihood function, 
 wherein the likelihood function comprises an expectation-maximization approach that alternates between computations of a first parameter and a second parameter where:
 the likelihood function takes the first parameter and the second parameter as given, and computes the expectation of the likelihood of the purchase of the product during the event conditional on the first parameter and the second parameter; and 
 a maximization is performed by taking latest values of the likelihood of the purchase of the product during the event, and solves for the first parameter and the second parameter while holding fixed the latest values, 
 
 further comprising identifying a critical set of products using the likelihood function for the event that causes a purchase model of a latent event response to deviate from a baseline purchase model, 
 wherein the likelihood of the event-driven purchase is estimated by maximizing the likelihood function, 
 further comprising iteratively updating a customer-event map and a purchase model for the event-driven purchase to create a model of latent event response. 
 
     
     
       2. A computer program product for market event identification, the computer program product comprising a non-transitory computer-readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform:
 defining a likelihood function that maps a likelihood of a purchase of a product during an event by customers having similar customer features, the function including a variable that factors event-product relationships; 
 estimating a likelihood of an event-driven purchase by the customers for each of a plurality of product-event combinations using the defined likelihood function, 
 wherein the likelihood function comprises an expectation-maximization approach that alternates between computations of a first parameter and a second parameter where:
 the likelihood function takes the first parameter and the second parameter as given, and computes the expectation of the likelihood of the purchase of the product during the event conditional on the first parameter and the second parameter; and 
 a maximization is performed by taking latest values of the likelihood of the purchase of the product during the event, and solves for the first parameter and the second parameter while holding fixed the latest values, 
 
 further comprising identifying a critical set of products using the likelihood, function for the event that causes a purchase model of a latent event response to deviate from a baseline purchase model. 
 
     
     
       3. The computer program product of  claim 2 , wherein the likelihood of the event-driven purchase is estimated by maximizing the likelihood function. 
     
     
       4. The computer program product of  claim 2 , further comprising iteratively updating a customer-event map and a purchase model for the event-driven purchase to create a model of latent event response. 
     
     
       5. The computer program product of  claim 2 , further comprising iteratively updating a customer-event map and the purchase model for the event-driven purchase to create a model of latent event response. 
     
     
       6. A market event identification system, said system comprising:
 a processor; and 
 a memory, the memory storing instructions to cause the processor to perform:
 defining a likelihood function that maps a likelihood of a purchase of a product during an event by customers having similar customer features, the function including a variable that factors event-product relationships; 
 estimating a likelihood of an event-driven purchase by the customers for each of a plurality of product-event combinations using the defined likelihood function, 
 
 wherein the likelihood function comprises an expectation-maximization approach that alternates between computations of a first parameter and a second parameter where:
 the likelihood function takes the first parameter and the second parameter as given, and computes the expectation of the likelihood of the purchase of the product during the event conditional on the first parameter and the second parameter; and 
 a maximization is performed by taking latest values of the likelihood of the purchase of the product during the event, and solves for the first parameter and the second parameter while holding fixed the latest values, 
 
 further comprising identifying a critical set of products using the likelihood function for the event that causes a purchase model of a latent event response to deviate from a baseline purchase model. 
 
     
     
       7. The system of  claim 6 , wherein the likelihood of the event-driven purchase is estimated by maximizing the likelihood function. 
     
     
       8. The system of  claim 6 , embodied in a cloud-computing environment.